基于KPCA和BPNN的模拟加载系统油压信号识别技术研究

摘要

To solve the signal recognition of the simulated loading system, this paper proposes a recognition method based on Ker-nel Principal Component Analysis ( KPCA) feature extraction and BP Neural Network ( BPNN) . The KPCA is applied to the data ex-traction of the original samples and then, the BP neutral network pattern classifier is used to identify six different working states of the device. The test results verifies the effectiveness of the method above, and some useful references are proveded for the characteris-tic analysis and pattern recognition of the similar hydraulic pressure signals.%为解决模拟加载系统油压信号的识别问题，提出了一种基于核主元分析( KPCA)特征提取和BP神经网络( BPNN)相结合的模式识别方法。该方法首先采用KPCA对原始样本数据进行特征提取，然后采用BPNN构造模式分类器，对工作装置6种不同工作状态信号进行识别。实验结果验证了该方法的有效性，为同类液压系统的信号特征分析及模式识别提供了参考。